Improving Production Schedules in Textile Manufacturing...

Improving Production Schedulesin Textile Manufacturing

Omar Abboud, Ricky Abichandani, Ilker Eraslan, Arnav Tainwala

1.Background Info

Background Info

Analyzed real world data from Kazareen Textile Company (KTC), an apparel company that manufactures, sources, and retails various textile products.

KTC deploys orders according to a5-step production process:

1. Knitting2. Dyeing3. Cutting4. Sewing5. Packing

Each step corresponds to a different set of machines, processes, and technical requirements like dyes and measurements.

Background Info

Management receives the following items prior to production:

Order quantityDue dateSpecifications

MaterialsColorsDimensions

KTC purchases pre-processed synthetic materials from third party processors, and begins scheduling production of a garment according to an EDD algorithm

2.Methodology

General Assumptions

rj = 0

For jobs received between 1/16 and 3/12, all materials were preordered on 3/30. Therefore every job is assumed to release at the same time.

Virtual MachinesTo accommodate for the option of a job ending halfway through the day and then starting a new job on that same day, we added a “virtual machine” to each stage.

SpeedsWe take into account each job’s weight when calculating the number of days required to process a job, serving as a de facto processing time.

Intercession Times

A significant portion of makespan is made up of time taken between steps

-1 Intercession Times

DyeingApril 4th

Intercession-1

KnittingApril 4th

Dyeing begins on the same day knitting is completed

-2 Intercession Times

PackingIntercession-2

Sewing

Packing begins the day before sewing is completed

3.New Schedule Proposal

Shortest Processing Time

Continuously Reevaluate

As jobs complete, we reevaluate the SPT protocol to ensure that they are scheduled appropriately in the queue.

Shortest Processing Time

Effective Release DatesDepending on which jobs have been completed, we consider effective release dates that either place or delay each job’s entry to the next machine slot.

Machine 1

Machine 2

Machine 3

Processing Time: 24 DaysAn improvement of 1 day using SPT over EDD

4.Analysis and Algorithm Formulation

2 | rj | Cmax

Algorithm

2 || Cmax 2 | rj | Cmax 4 | rj | Cmax 2 | rj | Cmax

Algorithm Walkthrough

Processing LengthThe algorithm collapses each step with its successive intercession time for a total “process length.”

Same Processing TimeIf two jobs have the same processing time, we order by EDD

1. Jobs are scheduled successively according to SPT in knitting.2. At every point at which a job becomes available for dyeing, jobs are scheduled if possible

according to SPT - in the case that a job hasn’t been released from knitting, we move on to the next job and reevaluate at the next release date.

3. We repeat this process for cutting, sewing, and packing. The final schedule is presented.

Final Schedule

The first job begins on April 2nd

The last job ends on April 26th

Preemption

It is unlikely that a textile manufacturing operation can introduce preemption, as when dyes and materials are loaded into a machine the entire job must be completed as opposed to interrupted and then completed at a later date.

This prevents the schedule from being completely efficient, as we can see there is a sewing machine that is idle for four days due to this phenomenon.

Other Scheduling Inefficiencies

❖ A job’s overall lateness can be determined by its performance in a specific department: for example, many jobs with similar sewing, cutting, dyeing, and knitting times can be subject to a different packing time.

❖ This indicates that the order in which jobs are scheduled (by virtue of the intercession times phenomenon especially) is significant.

❖ Keeping a job in intercession might incur costs that are more serious than lateness - areas for future research

Thanks!Any questions?